4-D Seismic: When Should We Acquire the next Seismic Survey?

Sena, Armando R.¹; Stoffa, Paul L.¹; Sen, Mrinal K.¹; Jin, Long ²; Seif, Roustam ^1
1 Institute for Geophysics, University of Texas at Austin, Austin, TX.
² E&D, Shell, Houston, TX.

Analysis of time-lapse seismic data for reservoir characterization was proposed over three decades ago, but only during the last 10 to 15 years has reached widespread acceptance as a useful reservoir management tool. However, when applying this technology, reservoir engineers and geoscientists must decide when the next seismic survey should be acquired. This decision is not easy, and depends, in general, upon the lithology of the reservoir formation, geological structures, the type of hydrocarbon, and the production process being used. Joint inversion of time-lapse seismic and production data allows us to estimate the reservoir model that explains, in an optimal sense, both seismic and production data. However, these results can also help guide the decision of when to do the next seismic survey resulting in a more cost effective strategy. In this work, we use synthetic data generated using part of the SPE-10 clastic reservoir model, and a two phases (oil and water) reservoir simulator, subject to water injection. The results reveal some criteria about the optimal time that the next time-lapse seismic survey should be acquired. Seismic data acquired at the initial stage (prior to the start of production), well log data, and production data, obtained after the start of production, are the data available to predict the time that the next seismic data survey will be acquired. During this initial stage of the joint inversion, the seismic data only reflect the initial state of the reservoir. However, well log data can be used to build a statistic rock physics model that can be used for reservoir simulation. The production data are then incorporated as it become available. Our inversions suggest that, the availability of new seismic surveys at very early stages of production are not very effective in improving the inverted reservoir model (i.e., the sensitivity to these data is very small), and that the sensitivity is greater at or very close to the times when the production response at the wells (water or oil cut curves) start to change. When the next time-lapse seismic data set is acquired near to this time, the match of the inverted model and the production prediction are considerably improved.

AAPG Search and Discovery Article #90090©2009 AAPG Annual Convention and Exhibition, Denver, Colorado, June 7-10, 2009